Transfer of care is associated with longer unsuccessful resuscitations

Brief Report

Transfer of care is associated with longer unsuccessful resuscitationsB

Jared Strote MD, MS*, Pamela Kohler RN, MPH

Division of Emergency Medicine, Box 356123, University of Washington Medical Center, Seattle, WA 98195, USA

Received 2 April 2007; revised 24 April 2007; accepted 28 April 2007


Objective: Accepted guidelines define when to terminate unsuccessful resuscitations. We examined whether such resuscitations last longer for transported arrests in the field compared with those occurring in the emergency department (ED).

Methods: This was a retrospective study of patients who died in an urban, academic ED over 32 months starting from January 2001. Total length of resuscitation and the interval occurring in-ED were compared for arrests in the ED and transported arrests from the field.

Results: A total of 132 patients met the criteria, of whom 71 (53.8%) arrested in the field. Mean overall resuscitation times were longer for arrests occurring in the field (44 minutes; 95% confidence interval [CI], 39-48) compared with those in the ED (19 minutes; 95% CI, 16-22; P b .001). Mean resuscitation intervals occurring in the ED were no different for arrests occurring in the field (16 minutes; 95% CI, 13-19) than in the ED (19 minutes; 95% CI, 16-22; P N .05).

Conclusions: Unsuccessful resuscitations were longer and beyond Guideline recommendations when arrests occurred in the field and were transported. The interval of resuscitation that occurred in the ED was the same whether or not prehospital resuscitation occurred.

(C) 2008


There are well-established guidelines limiting the duration of unsuccessful resuscitations. For medical cardiopulmonary arrests, the American Heart Association recommends a maximum length of 25 to 30 minutes of continuous pulseless- ness before rescuers should stop their efforts [1-3]. Similarly, for both blunt and penetrating posttraumatic circulatory arrests, the

? Data from this article were presented at the 2006 Society for Academic Emergency Medicine Research Forum in San Francisco, Calif.

* Corresponding author. Tel.: +1 206 598 0103; fax: +1 206 598 4569.

E-mail address: [email protected] (J. Strote).

0735-6757/$ – see front matter (C) 2008 doi:10.1016/j.ajem.2007.04.032

National Association of Emergency Medical Services Physi- cians and the American College of Surgeons Committee on trauma guidelines recommend, in most cases, to stop after 15 minutes of unsuccessful resuscitation in the field [4].

Both medical [5-7] and trauma [8-13] guidelines are based on a large body of evidence suggesting that after these periods of unsuccessful resuscitation, the chances for meaningful recovery are small and the risks of continued resuscitation rapidly begin to outweigh the benefits.

Emergency departments (EDs) are unique in that they perform such resuscitations not only on patients who arrest in front of the provider but also on patients who arrive in the midst of a prehospital resuscitation. We sought to determine

whether the length of unsuccessful resuscitations is longer when an arrest occurs in the field and is transported to the ED compared with an arrest that both begins and ends in the ED.


Study design and setting

The study uses a retrospective cohort design. Data were collected at an urban, academic, level I trauma center and covered a 32-month period from January 1, 2001, to August 31, 2003.

The study ED is in an area served by a 2-tiered emergency medical services response system, dispatching fire depart- ment-based basic life support (BLS) and Advanced life support responders. The hospital also serves as a regional trauma center, receiving a large number of air- transported patients from around the region. Both air and ground transport teams have capabilities to terminate a resuscitation in the field with base station approval. There are no written guidelines for termination; decisions are made on a case-by-case basis.

Study population, protocol, and measures

Inclusion criteria were all patients who died in the ED after a resuscitation attempt. Exclusion criteria included patients for whom care was withdrawn secondary to Advance directives and patients who were noted to be hypothermic in the ED (b36.0?C) and were treated for this state (defined as having a second temperature check or documented use of warmed fluids through any route).

Patients who had died in the ED during the study period were identified by a review of daily record sheets and collected by the principal investigator. The ED and emergency medical services records included in each patient’s medical record were then accessed. Resuscitation interval data from prehospital resuscitations were only taken from prehospital records rather than from ED records. The data were abstracted and entered into a spreadsheet by a single medical student using a standardized collection tool. Data abstraction was monitored by the principal investigator for accuracy. The abstractor was not blinded to the study purpose.

The study was approved by the home institution’s human subjects division with a waiver of consent.

The dependent variables were as follows: (1) the total length of each resuscitation, from the last palpable pulse (or the time the patient was found in arrest if no pulse was ever palpated) until the time of death; and (2) the interval of each such resuscitation that occurred in the ED.

The primary independent variable was the location of terminal arrest (prehospital vs ED).

Adjustment variables were hypothesized a priori as being potentially different among patients in the 2

independent variable groups. They included arrest etiol- ogy (traumatic or medical), the initial rhythm of the terminal event (asystole, pulseless electrical activity), ventricular fibrillation/tachycardia (combined, given the small number of ventricular tachycardia patients), age quartiles (pediatric [b19], young adult [19-39], middle- aged [40-64], and older adult [N64]), race (white or nonwhite, given the small number of nonwhite patients), season (3-month intervals starting in July when physicians in training are at the lowest level of experience for their responsibility level), and type of transport (ground vs air). Additional variables for traumatic arrests included the type (blunt or penetrating) and the presence or absence of a thoracotomy procedure.

An additional analysis was performed to see what proportion of patients for each location of arrest had resuscitations that exceeded guidelines. This was performed separately for each etiology with different time cutoffs for medical (30 minutes) and trauma (15 minutes) patients, reflecting differences in guideline recommendations.

Data analysis

Pearson ?2 analyses were used to compare the popula- tions in each arrest location group. A multiple linear regression was used to generate mean differences in resuscitation intervals for each of 3 models: all patients, Medical patients only, and trauma patients only. In the all patients model, there was additional adjustment for arrest etiology (medical or trauma). In the trauma patient model, there was additional adjustment of trauma etiology (blunt vs penetrating) and thoracotomy.

Differences in the proportion of patients whose resuscita- tions exceeded guidelines were compared using Pearson ?2 analyses.

Because of the large number of patients for whom data were missing from the charts, we also performed a comparison of this group with the main group for the variables that were available (location of arrest, arrest etiology, age, sex, race, season, and transport type).

STATA 9 for Macintosh (StataCorp, College Station, Tex) was used for all statistical analyses.


During the study period, 229 patients died in the ED. Of these, 4 (1.8%) had hypothermia and 24 (10.5%) had care

withdrawn, leaving 201. Of these patients, 69 (34.3%) had incomplete documentation, making a resuscitation timeline impossible to establish. This left 132 (65.7%) patients in the study group.

Patients with incomplete documentation were not sig- nificantly different from the study group (P b .05) for any of the variables studied except for transport type; there was a

larger percentage of missing data from patients transported by ground (P = .10).

For the study group, the populations of those who arrested in the field and those who arrested in the ED were compared. The only significant difference was in arrest type: a significant majority of penetrating trauma patients arrested in the field (Table 1).

Mean total resuscitation times were longer if the arrest began in the field for both trauma and medical patients (Fig. 1). Adjusted differences in mean total Resuscitation duration, based on location, confirmed this difference to be significant (P b .001) when analyzed individually by arrest etiology or overall (Table 2).

In the ED only, mean resuscitation times were similar whether the arrest began in the field or in the ED for both

Table 1 Demographic comparison of those who arrested in the field vs in the ED

Fig. 1 Mean total resuscitations. Unadjusted mean total duration of resuscitation for arrests beginning in the field and those beginning in the ED with 95% confidence intervals. Data are grouped by medical arrests, trauma arrests, and all patients. CPR indicates cardiopulmonary resuscitation.

ED arrest (n = 61)

Field arrest (n = 71)


Arrest type



26 (42.6%)

26 (36.6%)


3 (4.9%)

19 (26.8%)


32 (52.5%)

26 (36.6%)




45 (73.8%)

55 (77.5%)


16 (26.2%)

16 (22.5%)




39 (63.9%)

49 (69.0%)


19 (31.2%)

14 (19.7%)


3 (4.9%)

8 (11.3%)

Age (y)



3 (4.9%)

7 (9.9%)


21 (34.4%)

26 (36.6%)


20 (32.8%)

22 (31.0%)


17 (27.9%)

16 (22.5%)




20 (32.8%)

18 (25.4%)


11 (18.0%)

11 (15.5%)


13 (21.3%)

19 (26.8%)


17 (27.9%)

23 (32.4%)




23 (37.7%)

21 (29.6%)


38 (62.3%)

50 (70.4%)




17 (27.9%)

14 (19.7%)

Ventricular fibrillation/

9 (14.8%)

9 (12.7%)


Pulseless electrical

20 (32.8%)

17 (23.9%)



15 (24.6%)

31 (43.7%)

Thoracotomy (trauma

n = 35

n = 45





19 (54.3%)

21 (46.7%)

Population Pearson ?2 comparison for arrests that began in the field and those that began in the ED with P values.

trauma and medical patients (Fig. 2). Based on location, the adjusted differences of the mean total resuscitation time occurring in the ED were not significantly different (P N .05) when analyzed individually by arrest etiology or overall (Table 3).

For both medical and trauma patients, the proportion of resuscitations that exceeded recommended guidelines was significantly larger in the group arresting in the field than in that arresting in the ED (Table 4).


In our study, unsuccessful resuscitations for transported arrests in the field were continued significantly longer than unsuccessful resuscitations that began in the ED were. Importantly, this meant that the average failed resuscitation length for field arrests exceeded recommended maximum durations, whereas those for ED arrests did not; and a

Table 2 Adjusted differences in mean total resuscitation

Adjusted mean difference (min)

95% confidence interval






Medical only




Trauma only




Adjusted differences in mean total duration of resuscitation: arrested in field – arrested in ED. Data are grouped for medical arrests, trauma arrests, and all patients.

Arrest began in field

Arrest began in ED






Resuscitation N30 min

21 (80.8%)

5 (19.2%)






Resuscitation N15 min

43 (95.6%)

16 (45.7%)


Arrests beginning in the field are compared with those that began in the ED. Data are grouped by arrest etiology.

Fig. 2 Mean ED resuscitations. Unadjusted mean duration of resuscitation occurring in the ED for arrests beginning in the field and those beginning in the ED with 95% confidence intervals. Data are grouped for medical arrests, trauma arrests, and all patients.

Table 4 Proportion of resuscitations exceeding guideline limits

significantly larger percentage of resuscitations exceeding guidelines occurred with field arrests. To examine the ques- tion of where this difference came from, we measured the specific interval of resuscitation performed in the ED for each group and found no difference. Viewed in another way, the specific interval of a resuscitation that was run by the emergency physician followed guidelines whether or not that extended the overall resuscitation beyond the guidelines. There are many possible explanations for this finding. It is possible that prehospital arrests that are transported represent a unique subset of patients with a higher likelihood of survival and one that warrants extraordinary resuscitation measures. As we did not examine all patients who were resuscitated in the field, this question cannot be addressed by the present study. Our patients in the comparison group, however, who began their terminal event in the ED, had comparable demographics, initial terminal rhythm, and most importantly, identical outcomes with those who arrested in the field. Furthermore, prior studies have looked into why certain patients are transported during ongoing resuscitation and found a wide variety of factors, among them many

nonmedical ones [14].

Table 3 Adjusted differences in mean in ED resuscitation

Adjusted mean difference (min)

95% confidence interval




-5.7 to 2.9


Medical only


-12.0 to 5.7


Trauma only


-8.5 to 1.0


Adjusted differences in mean duration of resuscitation occurring in the ED: arrested in field – arrested in ED. Data are grouped for medical arrests, trauma arrests, and all patients.

There also could be therapies or evaluative tools available in the ED but not in the field that could have accounted for the difference. We looked into one, thoracotomy in trauma patients, and found no effect on overall time. We did not evaluate for ultrasonography– commonly used to confirm the absence of Cardiac activity and which certainly could extend resuscitation time but likely not for more than 1 to 2 minutes.

Although guidelines ultimately leave individual decisions up to the practitioner, it is clear that in most cases, after a defined period of pulselessness is met, the resuscitation should be terminated–the costs of continuing quickly begin to outweigh the benefits. It seems extremely unlikely that a special subgroup of patients or specific ED therapies could explain the large discrepancy from guidelines found in our group of transported patients but not in the other.

A different possible explanation is that field assessments may not always be considered reliable by the accepting EP. A recent study of trauma arrests by Pickens et al [15] provided examples of paramedic pulse assessments that did not correlate with initial ED evaluations. The clinical signifi- cance of this finding is unclear, but such studies may lead to clinicians erring on the side of distrusting prehospital reports. In our study, terminal events were only considered to have begun in the field if patients were documented as pulseless at last check by paramedics and never having a documented pulse in the ED; in these cases, the only question for the EP would be the paramedics’ accuracy in documenting the exact time the pulse was last lost.

Another way to account for the findings may be paramedic–EP communications. Prior studies have shown that EPs retain only 30% to 40% of paramedic report information in trauma cases [16]. It is possible that EPs are simply not registering how long a period of pulselessness has elapsed, focusing solely on the patient’s critical condition at the moment. This warrants further study; it is possible that with interventions to assure key communica- tion points, resuscitations may be terminated earlier after arrival in the ED.

If EPs are aware that they are extending resuscitations beyond accepted guideline limits, there are possible explanations for these decisions as well. One is a fear of

litigation. An analysis of patients presenting to the ED in asystole has been addressed in a prior article [17], suggesting protection for the EP based on a large literature demonstrating futility; one could easily extend this reason- ing to cover all patients who have met guideline limits for continued pulselessness.

Perhaps the most likely explanation of all for extended resuscitation is to display a continued heroic effort for the medics who began the resuscitation or for family who accompanied the patient to the ED. It is very difficult indeed to have the one ED intervention be a declaration of death: public and hospital staff expectations for resuscita- tions generally and ED capabilities specifically are inappropriately high [18,19]; there is likely a sense among physicians that they are expected to have in-hospital therapies that are unavailable in the field; and discussions with grieving family members are certainly easier when it can be said that “we did everything we could.” These visceral benefits are likely powerful influences on resusci- tation termination. In academic centers there is an additional pressure to continue resuscitations for the educational benefit, which has been argued as ethically justified under certain circumstances [20].

Although much less readily apparent in the highly charged environment of an ongoing resuscitation, there are costs to patient, caregivers, and society at large when resuscitations are inappropriately extended. Patients who undergo extended resuscitations and survive are at a much higher risk for Poor neurologic outcome [21]. Resuscitations frequently involve a higher risk to health care workers for exposure to high-risk body fluids [22-24]. Monetary costs to patients with extended resuscitations are higher [14]. And extended ED resuscitations detract from others access to care via both caregiver time and societal resources [14,25].

A great deal has been written about the ethics of inappropriately extended resuscitations [26,27]. The com- plex decision to end a resuscitation requires weighing many different costs and benefits but always hinges on the futility of continuing. In the ED, the decision can be further complicated by the transfer of care from one provider to another. In these cases, the use of prehospital resuscitation information should play an important role.


The current study is limited by its retrospective design and can therefore only show correlation, not causation. We were further limited by the number of charts that lacked a timeline of the resuscitation. It is possible that the subset of patients who had missing timelines differed in some significant way from the population we studied, skewing our results, although our additional analysis of the 2 groups did not show a difference. In addition, of those patients for whom we did collect data, there was a trend toward a larger

number of missing initial rhythm among the patients who arrested in the field; if the actual rhythms of this group significantly skewed the data, it could affect our conclu- sions. It is also possible that there are relevant differences that we did not measure in the populations of patients who arrest in the field and those who arrest in the ED, which may have changed our findings. Finally, our data come from one medical center and cannot be generalized to all EDs and emergency medical services systems.


The authors appreciate the data collection help from Dan Rogers, MD, and analysis from Lara Wagner, MD. They also appreciate the editorial help and guidance from Drs Carin Olson and Richard Cummins and the access to data provided by Dr Michael Copass.


  1. Eckstein M. Termination of Resuscitative efforts: medical futility for the trauma patient. Curr Opin Crit Care 2001;7(6):450-4.
  2. Bailey ED, Wydro GC, Cone DC. Termination of resuscitation in the prehospital setting for adult patients suffering nonTraumatic cardiac arrest. National Association of EMS Physicians Standards and Clinical Practice Committee. Prehosp Emerg Care 2000;4(2):190-5.
  3. Guidelines 2000 for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care. Part 6: advanced cardiovascular life support: section 7: algorithm approach to ACLS emergencies: section 7A: principles and practice of ACLS. The American Heart Association in collaboration with the International Liaison Committee on Resuscita- tion. Circulation 2000;102(8 Suppl):I136-9.
  4. Hopson LR, Hirsh E, Delgado J, et al. Guidelines for withholding or termination of resuscitation in prehospital traumatic cardiopulmonary arrest. J Am Coll Surg 2003;196(3):475-81.
  5. Smith JP, Bodai BI. Guidelines for discontinuing prehospital CPR in the emergency department–a review. Ann Emerg Med 1985;14 (11):1093-8.
  6. Eliastam M, Duralde T, Martinez F, Schwartz D. Cardiac arrest in the Emergency Medical Service System: guidelines for resuscitation. JACEP 1977;6(12):525-9.
  7. Bonnin MJ, Pepe PE, Kimball KT, Clark Jr PS. Distinct criteria for termination of resuscitation in the out-of-hospital setting. JAMA 1993;270(12):1457-62.
  8. Stratton SJ, Brickett K, Crammer T. Prehospital pulseless, unconscious penetrating Trauma victims: field assessments associated with survival. J Trauma 1998;45(1):96-100.
  9. Shimazu S, Shatney CH. outcomes of trauma patients with no vital signs on hospital admission. J Trauma 1983;23(3):213-6.
  10. Rosemurgy AS, Norris PA, Olson SM, Hurst JM, Albrink MH. Prehospital traumatic cardiac arrest: the cost of futility. J Trauma 1993;35(3):468-73 [discussion 473-464].
  11. Pasquale MD, Rhodes M, Cipolle MD, Hanley T, Wasser T. Defining “dead on arrival”: impact on a level I trauma center. J Trauma 1996;41 (4):726-30.
  12. Fulton RL, Voigt WJ, Hilakos AS. Confusion surrounding the treatment of traumatic cardiac arrest. J Am Coll Surg 1995;181(3):209-14.
  13. Battistella FD, Nugent W, Owings JT, Anderson JT. field triage of the pulseless trauma patient. Arch Surg 1999;134(7):742-5 [discussion 745-6].
  14. Hick JL, Mahoney BD, Lappe M. Factors influencing hospital transport of patients in continuing cardiac arrest. Ann Emerg Med 1998;32(1):19-25.
  15. Pickens JJ, Copass MK, Bulger EM. Trauma patients receiving CPR: predictors of survival. J Trauma 2005;58(5):951-8.
  16. Scott LA, Brice JH, Baker CC, Shen P. An analysis of paramedic verbal reports to physicians in the emergency department trauma room. Prehosp Emerg Care 2003;7(2):247-51.
  17. Levy F, Kelen G. resuscitation attempts in asystolic patients: the legal tail wagging the dog? J Emerg Med 2006;30(2):223-6.
  18. Thorns AR, Ellershaw JE. A survey of nursing and medical staff views on the use of cardiopulmonary resuscitation in the hospice. Palliat Med 1999;13(3):225-32.
  19. Diem SJ, Lantos JD, Tulsky JA. Cardiopulmonary resuscitation on television. Miracles and misinformation. N Engl J Med 1996;334(24):1578-82.
  20. Guidelines for cardiopulmonary resuscitation and emergency cardiac care. Emergency Cardiac Care Committee and Subcommittees, American Heart Association. Part VIII. Ethical considerations in resuscitation. JAMA 1992;268(16):2282-8.
  21. Bonnin MJ, Swor RA. Outcomes in unsuccessful field resuscitation attempts. Ann Emerg Med 1989;18(5):507-12.
  22. Soderstrom CA, Furth PA, Glasser D, Dunning RW, Groseclose SL, Cowley RA. HIV infection rates in a trauma center treating predominantly rural blunt trauma victims. J Trauma 1989;29(11): 1526-30.
  23. Kelen GD, Fritz S, Qaquish B, et al. Substantial increase in human immunodeficiency virus (HIV-1) infection in critically ill emer- gency patients: 1986 and 1987 compared. Ann Emerg Med 1989;18(4):378-82.
  24. Baker JL, Kelen GD, Sivertson KT, Quinn TC. Unsuspected human immunodeficiency virus in critically ill emergency patients. JAMA 1987;257(19):2609-11.
  25. Suchard JR, Fenton FR, Powers RD. Medicare expenditures on unsuccessful out-of-hospital resuscitations. J Emerg Med 1999;17 (5):801-5.
  26. Levack P. Live and let die? A Structured approach to decision-making about resuscitation. Br J Anaesth 2002;89(5):683-6.
  27. Larkin GL. Termination of resuscitation: the art of clinical decision making. Curr Opin Crit Care 2002;8(3):224-9.

Leave a Reply

Your email address will not be published. Required fields are marked *